WhiteHatCyberus / Sign-Language-to-Text-Translator-Using-Opencv-Tensorflow-Object-detection-CNN-and-Machine-Learning

An application to translate American Sign Language to Text

Warning : Outdated.

Sign-Language-to-Text-Translator

We have used concepts of machine learning, computer vision, object detection, and cnn (convolutional neural networks)

Instructions

1. If you intend to create gestures or run the translator, its advisable to run firstpage.py (Main UI made using Tkinter modules) within "Main menu and creation of datasets"
   

2. We have trained the MNIST dataset CNN model with Tensorflow Object Detection API.

3. The obtained accuracy of trained model is 99.3%.

4. Note: This accuracy was obtained with 20 epochs

5. Feel free to reduce the number of epochs to 5 or 10, accuracy may be 94-96%, which is sufficient.

6. When you run the translator, make sure your background has minimal disturbances and noise.

Where can you reach me??

If you come across errors or blocks while utilising this source code or application, feel free to write to me in email

Connect with me :

Linkedin

Finally!!

Thank you to you, for utilising this code and application as reference or for educational purposes!!
Also I'm thankful for the team I had while creating this!!